Revision history for Perl extension HTML::TableParser.

0.38  Tue May 13 22:28:38 EDT 2008
        - no code changes -- CPAN distribution had bad META.yml file

0.37  Fri Sep 21 11:52:34 EDT 2007
        - incorrect setting of plans if Test::Pod::Coverage wasn't available
        - quiet inappropriate perlcritic warning

0.36  Thu Sep 20 09:30:35 EDT 2007
        - add version to Table.pm to fix PAUSE indexer error

0.35  Wed Sep 19 21:22:22 EDT 2007
        - fix test to work with Test::More 0.71
	- reorganize to a more modern layout
	- new pod & perlcritic tests

0.34  Sat Sep  7 01:25:01 EDT 2002
        - HTML::Parser v. 3.25 didn't handle croaks in callbacks
	  correctly. 3.26 does; Makefile.PL was updated to require it.

0.33  Tue Sep  3 14:15:13 EDT 2002
        - MANIFEST was out of date.  oops.

0.32  Tue Sep  3 11:32:20 EDT 2002
        - extra </table> tags were not handled properly. it now
	  croaks if it hits an extra </table> tag.

0.31  Sat Apr 20 01:56:28 EDT 2002
        - was using features of Test::More v0.32 and didn't
	  specify a version in Makefile.PL, causing problems
	  with earlier versions.  Ooops.

0.3   Wed Apr 17 11:26:31 EDT 2002
	- specification of ids and column names in matches has been
	  extended and simplified.  id and cols now may take
	  arrayrefs of things to match. colre is deprecated.

	- matches can be made using literals, subroutines and qr//
	  created regexps.  the latter two are automatically recognized.

	- explicit exclusions and skipping of tables is now possible

	- improved docs (a bit)

	- decode_entitites was being called too late in fix_texts,
	  so Trim was less than effectual.

	- added DecodeNBSP attribute to fix nbsp decoding issues.

	- callbacks for methods and classes can be turned off
	  by assigning an undef value to the appropriate key in the
	  table request.  

	- the warn callback is also passed the line number now.
	  this is an incompatible change in the API.

0.2   Wed Jan 30 19:54:14 EST 2002

        - one could not use id=DEFAULT as a fall through if a column 
          name match didn't work.  in fact, one couldn't intersperse
          id's in a request list; they all had to be at the beginning.
          there's now a well defined order in how requests are
          processed.

	- fixing this uncovered another subtle bug.  a header line was
	  originally recognized only after finishing the first
	  non-header line.  this lead to the situation that if an
	  embedded table was in the first non-header row, it would be
	  matched against column name match requests before the enclosing
	  table, because it would be completely parsed before the
	  enclosing table's header was even recognized.  to fix this,
	  we finish off a header in the first non-header column if the
	  previous row was a header and the current row isn't a
	  header.
	
	- added an embedded table test in t/contents.t

	- format of comparison data was wrong. wanted embedded tab
	  between columns, got character sequence '\t'
	
	- checking whether a request had been used was 99% foobared.


0.1   Wed Dec 12 17:12:22 EST 2001
        - total rewrite.  the only thing left of the old
	code is probably the header/row column&row spanning code.

0.01  Fri Dec 10 14:36:39 1999
	- original version; created by h2xs 1.18